Kwi-Il Park , Minbaek Lee , Ying Liu , San Moon , Geon-Tae Hwang , Guang Zhu , Ji Eun Kim , Sang Ouk Kim , Do Kyung Kim , Zhong Lin Wang , and Keon Jae Lee *
Flexible Nanocomposite Generator Made of BaTiO 3 Nanoparticles and Graphitic Carbons
N
While global energy consumption has steadily increased in the past decades due to industrialization and population growth, [ 1 ] society is facing a problem with the depletion of fossil energy resources as well as environmental problems (such as global warming, carbon dioxide emissions, and damage to the ozone layer). [ 2 ] These challenges can be addressed by renewable energy resources, which are always available everywhere. [ 1 , 2 ] Outdoor renewable energy sources such as solar energy (15 000 μ W/cm 3 ), [ 3 , 4 ] wind energy (380 μ W/cm 3 ), [ 5 ] and wave energy (1 000 W/cm of wave crest length) [ 6 , 7 ] can provide large-scale needs of power. However, for driving small electronics in indoor or concealed environments [ 3 , 8 ] (such as in tunnels, clothes, and artifi cial skin) and implantable biomedical devices, innovative approaches have to be developed.
One way of energy harvesting without such restraints is to utilize piezoelectric materials that can convert vibrational and mechanical energy sources from human activities such as pres-sure, bending, and stretching motions into electrical energy. [ 9–11 ] Wang and co-workers [ 9 , 10 , 12–15 ] have used piezoelectric ZnO nanowire arrays to develop a nanogenerator technologies, who have demonstrated the feasibility using this type of generator to power commercial light-emitting diodes (LEDs), [ 13 ] liquid crystal displays, [ 14 ] and wireless data transmission. [ 15 ] These nanogen-erators can also convert tiny bits of biomechanical energy (from sources such as the movement of the diaphragm, the relaxa-tion and contraction of muscle, heartbeat, and the circulation of blood) into power sources. [ 16 , 17 ]
Recently, there have been attempts to fabricate thin fi lm-type nanogenerators [ 11 , 18 ] with perovskite ceramic materials (PbZr x Ti 1-x O 3 and BaTiO 3 ), which have a high level of inherent piezoelectric properties. The BaTiO 3 thin fi lm nanogen-erator has demonstrated by the authors [ 11 ] using the transfer process [ 19–22 ] of high temperature annealed perovskite thin fi lm from bulk substrates onto fl exible substrates; it generates a
K.-I. Park , S. Moon , G.-T. Hwang , J. E. Kim , Prof. S. O. Kim , Prof. D. K. Kim , Prof. K. J. Lee Department of Materials Science and EngineeringKorea Advanced Institute of Science and Technology (KAIST)291 Daehak-ro, Yuseong-gu, Daejeon, 305-701Republic of Korea E-mail: [email protected] Dr. M. Lee , Y. Liu , G. Zhu , Prof. Z. L. Wang School of Materials Science and EngineeringGeorgia Institute of Technology771 Ferst drive, Atlanta, Georgia, 30332-0245, USA
Adv. Mater. 2012, DOI: 10.1002/adma.201200105
much higher level of power density than other devices with a similar structure. [ 10 ]
Herein, we report the nanocomposite generator (NCG) achieving a simple, low-cost, and large area fabrication based on BaTiO 3 nanoparticles (NPs) synthesized via a hydrothermal reaction (see Method S1) [ 23 ] and graphitic carbons, such as single-walled and multi-walled carbon nanotubes (SW/MW-CNTs), and reduced graphene oxide (RGO). The BaTiO 3 NPs and carbon nanomaterials are dispersed in polydimethylsi-loxane (PDMS) by mechanical agitation to produce a piezoelec-tric nanocomposite (p-NC). The p-NC is spin-casted onto metal-coated plastic substrates and cured in an oven. Under periodic external mechanical deformation by bending stage or biome-chanical movements from fi nger/feet of human body, electric signals are repeatedly generated from the NCG device and used to operate a commercial red LED.
The schematic diagrams of the fabrication process are shown in Figure 1 a and detail information described in Experimental section. Figures 1 b and c show cross-sectional scanning elec-tron microscopy (SEM) images of a 250 μ m thick p-NC that is sandwiched between the top and bottom metal-coated plastic substrates. A magnifi ed cross-sectional SEM image (Figure 1 c) shows that the BaTiO 3 NPs and the MW-CNTs are well distrib-uted in the PDMS matrix. The BaTiO 3 NPs generate piezoelec-tric potential under external stress and act as an energy genera-tion source. The CNT’s role in an NCG device as dispersant, stress reinforcing agent, and conducting functional material is explained later. The volume percentage of CNT is controlled so low so that they do not affect the dielectric property of the entire structure. The hydrothermal BaTiO 3 NPs have a rounded shape with size of 100 nm (see Figure 1 d and Figure S1a-i). Raman analysis is conducted to provide a more comprehensive phase characterization of the BaTiO 3 NPs (the inset of Figure 1 d). The sharpness of the spectrum in the range of 305 to 720 cm − 1 is ascribed to the A 1 and E (longitudinal optical) modes, which are specifi c to a tetragonal phase of BaTiO 3 . [ 24 ] The crystalline structure of the BaTiO 3 NPs is characterized by X-ray diffrac-tion (see Figure S1a-ii). The Raman shift and X-ray diffraction pattern indicate that the BaTiO 3 NPs have good crystallinity with an excellent ferroelectric tetragonal phase. The MW-CNTs (Carbon Nano-material Tech. Co.), which are prepared via cata-lyst chemical vapor deposition, have a diameter of 5 to 20 nm and a length of ∼ 10 μ m (see TEM images of Figure S1b-i). We also employed the SW-CNTs with a diameter of ∼ 3 nm and RGO obtained by modifi ed Hummers methods [ 25 ] (see Method S2) to confi rm the universality of carbon nanomaterials in energy generation behavior (see Figures S1c and S1d). The pho-tograph in Figure 1 f shows the core p-NC material stretched by
Figure 1 . (a) Schematic illustration of the process for fabricating NCG device. (b) A cross-sectional SEM image of an NCG device. (c) A magnifi ed cross-sectional SEM photograph of the p-NC. (d) A SEM image of the BaTiO 3 NPs synthesized by hydrothermal method. The inset shows a Raman spectrum obtained from and BaTiO 3 NPs. (e) The MW-CNTs have a diameter of 20 nm and a length of 2 μ m. The inset shows a typical Raman shift of the MW-CNTs with large D bands. (f) Photograph of the p-NC stretched by tweezers. The inset shows the NCG device (3 cm x 4 cm) bent by fi ngers. (g) A large-area type NCG device (13 cm x 13 cm) fabri-cated by spin-casting or Mylar bar-coating.
tweezers; the inset shows a completely bended NCG device on metal-coated fl exible substrates. These images confi rm that the nanocomposites that generate self-powered energy are very fl ex-ible, bendable, and even stretchable (see Video S1). A large area NCG device (NCG pad, 13 cm x 13 cm) can also be fabricated by a process of spin-casting or Mylar bar-coating [ 26 ] (Figure 1 g). We believe that this arbitrary scalability to a large area fabrica-tion is a unique advantages over previous nanogenerator tech-niques [ 10 ] because it overcomes the size limitation caused by the high vacuum chamber growth. The thickness can also be easily adjusted by multiple spin-casting or simple die casting (see Figures S2 and S3).
We measured the generated outputs of the NCG device during the periodic bending and unbending motion of the bending
stage; the results are shown in Figure 2 . To exclude any artifact from the external elec-trostatic charges, the sample is placed in a Faraday cage on an optical table and bending system is well grounded. The measured output voltage and current (Figure 2 b) corre-spond to the three states of the NCG device shown in Figure 2 a: that is, the original state, the bending state, and the release state. To verify that the measured output signals are purely generated by the NCG sample, we con-duct a widely used switching-polarity test. [ 11 , 12 ] When a measurement instrument is forward connected to the device (corresponding to the upper inset of Figure 2 b-i), the NCG device generates a positive voltage and current upon the bending states (Figures 2 b-i and ii). In the case of the reverse connection (to the upper inset of Figure 2 c-i), the negative output pulses are measured. The results indi-cate that the measured outputs are the true signals generated from our NCGs strained by bending motions. Under the continual bending and unbending cycles, the NCG device repeatedly generates an open-circuit voltage (V oc ) of ∼ 3.2 V and a short-circuit cur-rent (I sc ) signal of 250 to 350 nA; these output values are produced for a maximum hori-zontal displacement of 5 mm from an orig-inal 4 cm long sample at a deformation rate of 0.2 m ⋅ s − 1 (see Video S1 for real-time live views of the energy harvesting). The ampli-tude of the output voltage generated from the NCG device increases from 0.2 V to 3.2 V after the poling process (see Figure S4), and also depends on the composition of nanoma-terials (Figure S5 and Table S1), the angular bending strain (Figure S6), and strain rate (Figure S7). The durability test is conducted to confi rm the mechanical stability of the NCG device (Figure 2 d). The voltage amplitude does not appear to have changed signifi cantly after 1200 bending cycles (more than 1 hour of operation). The NCG device also shows superior mechanical stability during 600
bending cycles after 3 and 7 days of testing. The bottom insets of Figure 2 d show the magnifi ed output signals (see Figure S8 for the durability test result of the current signals).
The detail power generation mechanism of the NCG device is described in Figure S9. In addition, we calculate the piezopo-tential distributions inside the NCG by using a simple rectan-gular model composed of six BaTiO 3 NPs in a PDMS matrix (see Method S3) as shown in Figure 3 a and b. The material parameters of the BaTiO 3 and PDMS, which are taken from the COMSOL simulation software, are used for the fi nite ele-ment analysis. [ 27 ] When the three-layer structure comprised of a top thin plastic substrate, a layer of six BaTiO 3 NPs-embedded PDMS, and a bottom thick plastic substrate is bent, the tensile stress is built-up over the entire PDMS matrix due to the shift
heim Adv. Mater. 2012, DOI: 10.1002/adma.201200105
www.advmat.dewww.MaterialsViews.com
CO
MM
UN
ICATIO
N
Figure 2 . (a) Optical images of NCG devices in their original, bending, and release states. (b) The measured output voltage and current signals of the NCG device in the forward con-nection during the periodic bending and unbending motions. (c) The open-circuit voltage and short-circuit current signals generated in the reverse connection. (d) The durability test results conducted to confi rm the mechanical stability of the NCG device.
of the neutral plane to bottom side. Figure 3 b shows the color-coded piezopotential distribution of the PDMS between the top and bottom electrodes. The calculated piezopotential for tensile strain of 0.33% in the X-axis is generated across the top and bottom sides of the PDMS matrix by the piezoelectric NPs.
We characterized the NCG device without any CNTs in order to exploit the role of the CNTs within the device. The charac-terization is supported by the calculated piezopotential and measurement results ( Figures 4 a and b). It is a challenging task to disperse BaTiO 3 NPs uniformly inside the composite for a high-performance NCG device. Only the BaTiO 3 NPs in the PDMS matrix cannot avoid the aggregation and poor
dispersion (the schematics of Figure 4 a-i); thus, they lead to the low output voltage (Figure 4 b-i). In the case of the NCG devices with carbon nanomaterials (Figures 4 a-ii and 1 c), the BaTiO 3 NPs can be well dis-tributed by forming a complex mixture with CNT networks; as a results, a high output voltage is generated (Figure 4 b-ii). The above theories are well supported by the compara-tive results of the measured output voltage of devices with and without CNT networks (Figures 4 b-i and ii). We also calculated the piezopotential distribution of the aggregated NP model (Figure 4 a-iii). The piezopotential of the aggregated NP model is inferior to that of a well-distributed NP model (Figures 3 a and b) that is consistent with our assertions. Another role of CNT is to reinforce the stress applied to NPs by enhancing the composite stress; the enhancement is attributed to a change in the mechanical property of the composites material. [ 28 , 29 ] Since the CNTs are well mixed and entangled with NPs in PDMS matrix, the small NPs can signifi cantly be stressed. This phenomenon is depicted in the schematic illustrations of Figures 4 a-iv and 4 a-v. The calculation of the piezopoten-tial distribution of the CNT-reinforced model is based on the modifi ed parameters [ 30 ] of the PDMS that is affected by the addition of CNTs (Figure 4 a-vi). In accordance with our initial assumption, the piezopotential differ-ence of the reinforced model is higher than that of the non-CNT model (Figures 3 and b). Finally, the conduction paths formed by the CNT networks can reduce the internal resistance of NCG devices [ 31 ] and thus the resistance, yielding the short voltage lifetime and the high output characteristics. [ 32 , 33 ] As shown in the bottom-right insets of Figures 4 b-i and ii, the NCG device generates a sharper voltage peak compared to the device with only NPs. The equivalent circuits of the NCG device are depicted in the bottom-left inset of Figure 4 b-i, where R N , C N , and R L are the internal NCG resistance, the NCG capaci-tance, and the load resistance of the meas-
urement system (M/S), respectively. The piezoelectric potentials are generated by the mechanical deformation and subsequently removed during the RC discharging process with a time con-stant ( τ ) according to the equation (1) . [ 32 ]
τ = (RN + RL ) · CN (1)
The NCG device shows a shorter output voltage lifetime (50.29 ms), which is defi ned as full width at half maximum (FWHM) of the voltage peak, than that (71.51 ms) of the device without CNTs (see the Figure S10 and Table S1 for the details of voltage lifetime calculated from the voltage-time plots). It indicates that a short lifetime of NCG device caused by CNT
3wileyonlinelibrary.comeim
www.advmat.dewww.MaterialsViews.com
CO
MM
UN
ICATI
ON
Figure 3 . (a) Simulation model of a NCG device. The entire structure includes the six BaTiO 3 NPs-embedded PDMS between the top and bottom plastic substrate. (b) The simulated piezo-potential difference inside the PDMS between top and bottom electrodes indicated by color code.
additions leads to increasing the output voltage. The PDMS sample with only MW-CNTs is also characterized and compared to other devices to confi rm that the measured outputs are gen-erated from piezoelectric NPs (see Figure S11a). No reliable signals are observed upon bending and unbending motions. This result indicates that the output signals of NCG device are
4
Figure 4 . (a) Schematics of the cross-sectional structure of NCG devices and the calculated piezopotential distributions for explaining the role of CNTs. The CNTs act as dispersing (ii) and stress reinforcing agents (v) which are well supported by the calculated piezopotential differ-ence (iii and vi). (b) The output voltages generated from a device containing only BaTiO 3 NPs and a NCG device. The good distribution of NPs and stress enhancement by CNTs lead to the superior output voltage. The bottom-right insets show the magnifi ed output voltage generated by the mechanical bending motion. The bottom-left inset of shows the equivalent circuit of NCG devices.
obtained from charges generated by piezo-electric NPs.
The NCG devices with SW-CNTs and RGO are fabricated and characterized to study the electric and geometric effects of other graph-itic carbon nanomaterials instead of the MW-CNTs (see Figure S12). The NCG device that contains the nanomaterials of the SW-CNTs and BaTiO 3 NPs generates an output voltage of ∼ 3.2 V, which is similar to that of the MW-CNT-based NCG device shown in Figure 4 b-ii. The similarity indicates that the electrical properties of CNT have an insignifi cant infl u-ence on outputs of an NCG device. Unlike the high performance of MW/SW-CNT-based devices, the NCG device with RGO produces a lower output voltage ( ∼ 2.0 V) than a CNT-based NCG device. This distinction is likely due to the difference in the degree of mixing, which in turn is caused by the geometrical difference between the CNT networks and the laminated RGO structures.
We demonstrated the energy harvesting which converts human muscle movement into electrical energy (Figure S13). The small-scaled device and NCG pad are driven by the slight bending motions of fi nger tapping and regular foot stepping, respectively. The voltage and current produced by the slow fi nger motion reaches up to ∼ 200 mV and ∼ 7 nA, respectively (see Figure S13b and Video S1); the output values by fi nger tapping are lower than those of bending stage with high speed deformation of Figure 2 . The NCG pad repeat-edly generates the output signals of ∼ 1.5 V
and ∼ 150 nA (see Figure S13d and Video S1); these output values are produced from an activation area of 5 cm x 7 cm with applying force of ∼ 200 N (i.e., ∼ 57 kPa).
For a potential utilization of our energy harvesting technology, we demonstrated the lit up of a commercial LED solely with the electricity generated from the NCG device. As shown in Figure S14a, the NCG device is connected to a predesigned circuit composed of diodes and capacitors; it is activated by bending and unbending motions for ∼ 1 hr (2.5 Hz) to accomplish the energy storage process. The alternating energy sources are rectifi ed (Figures S14b and S14c) and simultaneously stored in the four capacitors in parallel. The voltage charged to a single capacitor is ∼ 0.6 V, which is lower than the
peak output voltage of NCG devices due to voltage drop drained by the rectifying circuits or the capacitor leakage. [ 13 ] The total output voltage of the four charged capacitors aligned in serial alignment reaches up to ∼ 2.37 V, which is suffi cient to drive a commercial LED. Figure 5 a shows the captured image of a red LED integrated into the circuit shown in the top inset. When
Figure 5 . (a) A captured image of a commercial red LED incorporated into the circuit. The bottom inset shows the measured voltage ( ∼ 2.37 V) when the four capacitors are aligned in a serial. (b) A captured photograph showing an LED lit up by the electric energy generated from an NCG device.
the switch is turned on (the inset of Figure 5 b), the LED is lit up and lasted for ∼ 3 s (Figure 5 b) (see Video S1 for a video clip of the LED being lit up by the electricity generated from the NCG device). The commercial microelectronic LED is success-fully operated by the piezoelectric energy from the NCG device without any external electric power source.
In summary, we have fabricated the NCG device based on piezoelectric BaTiO 3 NPs and universal graphitic carbons. The BaTiO 3 NPs are mixed with one of carbon nanomaterials (SW/MW-CNTs and RGO) to prepare the p-NC. The NCG devices obtained by spin-casting/bar-coating method convert the mechanical deformation and even tiny biomechanical move-ments into electric energy. We have theoretically analyzed the principle of power generation and CNT’s role in NCG device, which have been supported by the fi nite element simulation. The alternating energy generated from an NCG device is stored in capacitors and subsequently used to turn on a commer-cial LED device. Our p-NC technique successfully overcomes the size-related restrictions existed in previous nanogenerator and enables simple, low-cost, and large-scale self-powered energy system. This result innovatively expands the feasibility of self-powered energy systems for application in consumer electronics, sensor networks, and energy harvesting in indoor environments.
Experimental Section Spin-casting of the dielectric layer onto the metal-coated plastic substrates :
Radio frequency magnetron sputtering is used to deposit layers of Cr (10 nm) and Au (100 nm) onto fl exible substrates (Kapton fi lm, 25 μ m and 125 μ m in thickness) for the top and bottom electrodes. A layer of PDMS (Sylgard 184, Dow Corning) is spin-casted onto Au/Cr/plastic substrates to form a dielectric layer and then cured at 85 ºC for 10 min in an oven.
Production of the p-NC : The BaTiO 3 NPs are prepared via hydrothermal reactions at 200 ° C for 24 hr. They are mixed with a graphitic carbon (either SW/MW-CNTs or RGO) in specifi c proportions (the compositional details of which are shown in Table S1 of the Supporting Information). The mixture is then stirred for ∼ 5 hr in ethanol with a magnetic bar. After the subsequent drying and granulation, the well-mixed nanomaterials (compositions of 1 wt% MW-CNTs and 12 wt% BaTiO 3 NPs) are poured into a PDMS matrix for the fi nal p-NC product (see the Supporting
Information for a video clip on the production process of p-NC, Video S1).
Spin-casting of the p-NC layer : The p-NC is spin-casted onto PDMS/Au/Cr/plastic substrates at a spinning rate of 1500 rpm for 30 sec and cured at 85 ºC for 5 min in an oven.
Fabrication of the NCG : Another top PDMS/metal-coated plastic substrate is placed in uniform contact with a p-NC/PDMS/metal-coated bottom fl exible substrate and fully cured at room temperature for 1 day. Finally, the NCG device is poled at 150 ° C by applying an electric fi eld of 100 kV/cm for 20 hr. The fabricated NCG device maintains mechanical stability up to an extreme bending radius of 0.8 cm for many bending cycles. The Cu wires are attached to metal pads by means of silver (Ag) paste for the characterization of the output voltage and current signals.
Supporting Information Supporting Information is available from the Wiley Online Library or from the author.
Acknowledgements This work was supported by the Basic Science Research Program (grant code: 2011-0003483, CAFDC-2012-0000824) and the Smart IT Convergence System as Global Frontier Project (SIRC 2011-0031848) through the National Research Foundation of Korea (NRF). This research was also supported by a National Platform Technology (grant code: 10033707) funded by the Ministry of the Knowledge Economy of Korea. ZLW thanks the support from NSF, DARPA and BES DOE.
Received: January 9, 2012 Revised: March 12, 2012
Published online:
[ 1 ] G. J. Aubrecht , Energy: Physical, Environmental, and Social Impact , Pearson Education , London 2006 .
[ 2 ] C. Beggs , Energy: Management, Supply and Conservation , Elsevier , Oxford 2002 .
[ 3 ] S. Priya , D. J. Inman , Energy Harvesting Technologies , Springer Science , New York , 2009 .
[ 4 ] S. Roundy , P. K. Wright , J. Rabaey , J. Comput. Commun. 2003 , 26 , 1131 .
[ 5 ] S. Roundy , E. S. Leland , J. Baker , E. Carleton , E. Reilly , E. Lai , B. Otis , J. M. Rabaey , P. K. Wright , V. Sundararajan , IEEE Pervas. Comput. 2005 , 4 , 28 .
[ 6 ] K. Rhinefrank , presented at Energy Ocean 2005, April 26–28, Washington, DC .
[ 7 ] US Department of the Interior (Minerals Management Service), Technology White Paper on Wave Energy Potential on the US Outer Continental Shelf, 2006. (http://ocsenergy.anl.gov/documents/docs/OCS_EIS_WhitePaper_Wave.pdf).
[ 8 ] Y. Qi , M. C. McAlpine , Energy Environ. Sci. 2010 , 3 , 1275 . [ 9 ] Z. L. Wang , J. H. Song , Science 2006 , 312 , 242 . [ 10 ] S. Xu , Y. Qin , C. Xu , Y. G. Wei , R. S. Yang , Z. L. Wang , Nat.
Nanotechnol. 2010 , 5 , 366 . [ 11 ] K. I. Park , S. Xu , Y. Liu , G. T. Hwang , S. J. L. Kang , Z. L. Wang ,
K. J. Lee , Nano Lett. 2010 , 10 , 4939 .
5wileyonlinelibrary.comheim
www.advmat.dewww.MaterialsViews.com
CO
MM
UN
ICATI
ON
[ 12 ] R. S. Yang , Y. Qin , L. M. Dai , Z. L. Wang , Nat. Nanotechnol. 2009 ,
4 , 34 . [ 13 ] G. Zhu , R. Yang , S. Wang , Z. L. Wang , Nano Lett. 2010 , 10 , 3151 . [ 14 ] Y. Hu , Y. Zhang , C. Xu , G. Zhu , Z. L. Wang , Nano Lett. 2010 ,
10 , 5025 . [ 15 ] Y. Hu , Y. Zhang , C. Xu , L. Lin , R. L. Snyder , Z. L. Wang , Nano Lett.
2011 , 11 , 2572 . [ 16 ] R. Yang , Y. Qin , C. Li , G. Zhu , Z. L. Wang , Nano Lett. 2009 , 9 , 1201 . [ 17 ] Z. Li , G. A. Zhu , R. S. Yang , A. C. Wang , Z. L. Wang , Adv. Mater.
2010 , 22 , 2534 . [ 18 ] Y. Qi , J. Kim , T. D. Nguyen , B. Lisko , P. K. Purohit , M. C. McAlpine ,
Nano Lett. 2011 , 11 , 1331 . [ 19 ] K. I. Park , S. Y. Lee , S. Kim , J. Chang , S. J. L. Kang , K. J. Lee , Electro-
chem. Solid-State Lett. 2010 , 13 , G57 . [ 20 ] S. Kim , H. Y. Jeong , S. K. Kim , S.-Y. Choi , K. J. Lee , Nano Lett. 2011 ,
11 , 5438 . [ 21 ] S. Y. Lee , K. I. Park , C. Huh , M. Koo , H. G. Yoo , S. Kim , C. S. Ah ,
G. Y. Sung , K. J. Lee , Nano Energy 2011 , 1 , 145 . [ 22 ] M. Koo , S. Y. Park , K. J. Lee , Nanobiosensors in Disease Diagnosis
2012 , 1 , 5 . [ 23 ] D. R. Chen , X. L. Jiao , J. Am. Ceram. Soc. 2000 , 83 , 2637 .
[ 24 ] A. D. Li , C. Z. Ge , P. Lu , D. Wu , S. B. Xiong , N. B. Ming , Appl. Phys. Lett. 1997 , 70 , 1616 .
[ 25 ] J. E. Kim , T. H. Han , S. H. Lee , J. Y. Kim , C. W. Ahn , J. M. Yun , S. O. Kim , Angew. Chem. Int. Edit. 2011 , 50 , 3043 .
[ 26 ] K. J. Lee , M. J. Motala , M. A. Meitl , W. R. Childs , E. Menard , A. K. Shim , J. A. Rogers , R. G. Nuzzo , Adv. Mater. 2005 , 17 , 2332 .
[ 27 ] D. Armani , C. L. Liu , N. Aluru , Twelfth IEEE International Confer-ence on Micro Electro Mechanical Systems, January 17–21, 1999 , Orlando, USA , p. 222 .
[ 28 ] J. Suhr , N. Koratkar , P. Keblinski , P. Ajayan , Nat. Mater. 2005 , 4 , 134 .
[ 29 ] V. P. Veedu , A. Y. Cao , X. S. Li , K. G. Ma , C. Soldano , S. Kar , P. M. Ajayan , M. N. Ghasemi-Nejhad , Nat. Mater. 2006 , 5 , 457 .
[ 30 ] L. Ci , J. Suhr , V. Pushparaj , X. Zhang , P. M. Ajayan , Nano Lett. 2008 , 8 , 2762 .
[ 31 ] C. Seoul , Y. T. Kim , C. K. Baek , J. Polymer. Sci. Part B 2003 , 41 , 1572 . [ 32 ] C. Chang , Y.-K. Fuh , L. Lin , Solid-State Sensors, Actuators and
Microsystems Conference (Transducers), June 21–25, 2009 , Denver, USA , p. 1485 .
[ 33 ] C. E. Chang , V. H. Tran , J. B. Wang , Y. K. Fuh , L. W. Lin , Nano Lett. 2010 , 10 , 726 .
